Contrast-Induced Nephropathy: We Need All the Data to Discern the Truth

Contrast-induced nephropathy (CIN) usually is defined as an acute deterioration in kidney function after intravascular injection of iodinated contrast media. The entity has engendered a vast amount of research interest during the past 30 years, partly because it appeared to be a major cause of acute kidney injury in general1 and partly because as a form of acute kidney injury with a predictable time of injury, it is open to application of timely preventive measures. Most cases of CIN are characterized by transient and minor changes in markers of kidney function, such as serum creatinine level. As such, many would question the clinical importance of CIN. Dialysis as a result of CIN is required in less than 1% of cases, and in many, CIN is the final insult for people with already very low levels of kidney function. The other reason to continue to have concern about the more minor changes in kidney function seen in most patients with CIN is the association of these events with other more clearly serious adverse health outcomes, including major cardiovascular events, prolonged hospitalization, and early death. Such associations have been found consistently in many studies, but the reason for the association has not been clearly established. If CIN events are causally related to the later adverse outcomes, there would be a strong reason to try to identify effective preventive measures against CIN.

Despite the quantity of research devoted to identifying a means to prevent CIN, the entire field remains full of uncertainty. There are many reasons for this state of affairs. One may be the lack of clarity around the mechanism for the decreases in kidney function in humans with apparent CIN. Animal models generally require that additional insults to the kidney be combined with contrast exposure to reliably induce CIN.2 It is unclear whether the same apparent mechanisms of toxicity apply in humans as in these animal studies. The vast array of preventive options studied in clinical trials in humans to date has been based in part on presumed pathogenetic mechanisms. The approaches used might be grouped loosely into efforts to increase urine flow, use of drugs causing renal vasodilatation, and use of therapies aimed at reducing oxidative stress within the kidney. Each of these approaches has some merit given current concepts of the pathogenesis of CIN.2

Looking back at the literature on efforts to prevent CIN, a common trend seems apparent. Initial clinical trials of a particular approach are of limited size and report extremely positive results. Subsequently, more definitive studies either fail to replicate these findings or, when a body of trials has been completed, meta-analyses show heterogeneity which is hard to explain, and one is left with uncertainty about the effectiveness of a given approach. This has been the case, for example, for N-acetylcysteine3, 4 and isotonic sodium bicarbonate.5, 6 Current guidelines tend to emphasize the role of additional fluid in efforts to prevent CIN, although the optimal regimen is not clear.7 Given the apparent beneficial effect of fluid, there have been many studies in which diuretics are combined with fluid to further augment tubular fluid flow. In cases in which furosemide has been one of the diuretics used, the additional rationale has been that blocking medullary tubular sodium reabsorption should reduce tubular work and thus the requirement for oxygen at a time when contrast might be expected to cause a decrease in medullary oxygen delivery. However, forced diuresis approaches using furosemide generally have been associated with more rather than less frequent development of what has been labeled as CIN.8, 9 One criticism of some of these studies has been that the diuretic approach may have led to volume depletion and thus the increased rate of decrease in kidney function.

In this issue of the American Journal of Kidney Diseases, Majumdar et al10 report results of a randomized clinical trial in which they compared the efficacy of a forced diuretic approach using both mannitol and furosemide in combination with sufficient 0.45% saline to more than replace urinary losses with the use of 0.45% saline alone.10 As with many trials in this field, the study was of limited size (92 patients) and involved patients with existing kidney disease undergoing coronary angiography. A fairly standard definition of CIN was used: increase in serum creatinine level greater than 25% or 0.5 mg/dL greater than baseline within 48 hours after contrast. CIN as defined was observed in 50% undergoing forced diuresis compared with 28% given saline alone (adjusted odds ratio, 3.7; 95% confidence interval, 1.1 to 12.4). There was no difference between treatment groups in other clinical outcomes, including death and need for dialysis.

There are some methodologic issues with this trial. Although Majumdar et al10 used 0.45% saline intravenously and gave at least 500 mL before contrast administration, the exact state of hydration of each patient by the time of angiography is unknown, and a previous study suggested that 0.9% saline was more effective than 0.45% saline in preventing CIN.11 The investigators recruited less than half of their planned sample before the study was terminated. The rate of CIN in both groups was greater than expected and also greater than that in some prior studies in similar populations. Majumdar et al10 found that greater contrast dose was associated with greater risk of CIN, and the contrast dose was greater in the forced-diuresis group. Given all these data, the trial reported by Majumdar et al10 is not itself definitive. However, the study adds to the existing body of literature and decreases the level of uncertainty about whether forced diuretic approaches should be used for the prevention of CIN. At this time, such an approach clearly is not supported.

One intriguing issue with this trial is the 8-year lag from the end of recruitment to reporting. The exact reasons for this delay are not clear. In the present situation, I do not believe this renders the data “out of date,” and the investigators and the journal are commended for making these data public. It is very important that all data from clinical trials are reported because failure to report negative or null results clearly can lead to bias in the literature and ineffective therapies may continue to be used. Evidence of such possible publication bias has been seen recently in a meta-analysis of bicarbonate versus saline for the prevention of CIN.6 Into the future, the present drive to register clinical trials should allow additional assessment of biases caused by nonreporting. Interestingly the present trial was registered, but only 5 years after completion.

Uncertainty remains in relation to CIN. What we really need now is a big trial to assess whether an intervention not only prevents CIN, but also the associated major clinical end points. The difficult part may be deciding what intervention to test.